X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FDownloadImages.php;h=487a3a238eff43ff4316f1897f9fb65deb5310e2;hb=e6f12c073416eba6fc620f0ff38a343bda428280;hp=32a9dbb2ff723c40e468d74acbc63bd829e1f732;hpb=419214d7221e0821ef2b73eb2b3db816ed0cf173;p=github%2Fwallabag%2Fwallabag.git

diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
index 32a9dbb2..487a3a23 100644
--- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php
+++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
@@ -2,146 +2,318 @@
 
 namespace Wallabag\CoreBundle\Helper;
 
-use Psr\Log\LoggerInterface as Logger;
+use GuzzleHttp\Client;
+use GuzzleHttp\Message\Response;
+use Psr\Log\LoggerInterface;
 use Symfony\Component\DomCrawler\Crawler;
+use Symfony\Component\Finder\Finder;
+use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
 
-define('REGENERATE_PICTURES_QUALITY', 75);
+class DownloadImages
+{
+    const REGENERATE_PICTURES_QUALITY = 80;
 
-class DownloadImages {
-    private $folder;
-    private $url;
-    private $html;
-    private $fileName;
+    private $client;
+    private $baseFolder;
     private $logger;
+    private $mimeGuesser;
+    private $wallabagUrl;
 
-    public function __construct($html, $url, Logger $logger) {
-        $this->html = $html;
-        $this->url = $url;
-        $this->setFolder();
+    public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger)
+    {
+        $this->client = $client;
+        $this->baseFolder = $baseFolder;
+        $this->wallabagUrl = rtrim($wallabagUrl, '/');
         $this->logger = $logger;
-    }
+        $this->mimeGuesser = new MimeTypeExtensionGuesser();
 
-    public function setFolder($folder = "assets/images") {
-        // if folder doesn't exist, attempt to create one and store the folder name in property $folder
-        if(!file_exists($folder)) {
-            mkdir($folder);
-        }
-        $this->folder = $folder;
+        $this->setFolder();
     }
 
-    public function process() {
-        //instantiate the symfony DomCrawler Component
-        $crawler = new Crawler($this->html);
-        // create an array of all scrapped image links
-        $this->logger->log('debug', 'Finding images inside document');
-        $result = $crawler
-            ->filterXpath('//img')
-            ->extract(array('src'));
+    /**
+     * Process the html and extract image from it, save them to local and return the updated html.
+     *
+     * @param int    $entryId ID of the entry
+     * @param string $html
+     * @param string $url     Used as a base path for relative image and folder
+     *
+     * @return string
+     */
+    public function processHtml($entryId, $html, $url)
+    {
+        $crawler = new Crawler($html);
+        $imagesCrawler = $crawler
+            ->filterXpath('//img');
+        $imagesUrls = $imagesCrawler
+            ->extract(['src']);
+        $imagesSrcsetUrls = $this->getSrcsetUrls($imagesCrawler);
+        $imagesUrls = array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
+
+        $relativePath = $this->getRelativePath($entryId);
 
         // download and save the image to the folder
-        foreach ($result as $image) {
-            $file = file_get_contents($image);
+        foreach ($imagesUrls as $image) {
+            $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
+
+            if (false === $imagePath) {
+                continue;
+            }
 
-            // Checks
-            $absolute_path = self::getAbsoluteLink($image, $this->url);
-            $filename = basename(parse_url($absolute_path, PHP_URL_PATH));
-            $fullpath = $this->folder."/".$filename;
-            self::checks($file, $fullpath, $absolute_path);
-            $this->html = str_replace($image, $fullpath, $this->html);
+            // if image contains "&" and we can't find it in the html it might be because it's encoded as &
+            if (false !== stripos($image, '&') && false === stripos($html, $image)) {
+                $image = str_replace('&', '&', $image);
+            }
+
+            $html = str_replace($image, $imagePath, $html);
         }
 
-        return $this->html;
+        return $html;
     }
 
-    private function checks($rawdata, $fullpath, $absolute_path) {
-        $fullpath = urldecode($fullpath);
+    /**
+     * Process a single image:
+     *     - retrieve it
+     *     - re-saved it (for security reason)
+     *     - return the new local path.
+     *
+     * @param int    $entryId      ID of the entry
+     * @param string $imagePath    Path to the image to retrieve
+     * @param string $url          Url from where the image were found
+     * @param string $relativePath Relative local path to saved the image
+     *
+     * @return string Relative url to access the image from the web
+     */
+    public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
+    {
+        if (null === $imagePath) {
+            return false;
+        }
 
-        if (file_exists($fullpath)) {
-            unlink($fullpath);
+        if (null === $relativePath) {
+            $relativePath = $this->getRelativePath($entryId);
         }
 
-        // check extension
-        $this->logger->log('debug','Checking extension');
+        $this->logger->debug('DownloadImages: working on image: ' . $imagePath);
+
+        $folderPath = $this->baseFolder . '/' . $relativePath;
 
-        $file_ext = strrchr($fullpath, '.');
-        $whitelist = array('.jpg', '.jpeg', '.gif', '.png');
-        if (!(in_array($file_ext, $whitelist))) {
-            $this->logger->log('debug','processed image with not allowed extension. Skipping '.$fullpath);
+        // build image path
+        $absolutePath = $this->getAbsoluteLink($url, $imagePath);
+        if (false === $absolutePath) {
+            $this->logger->error('DownloadImages: Can not determine the absolute path for that image, skipping.');
 
             return false;
         }
 
-        // check headers
-        $this->logger->log('debug','Checking headers');
-        $imageinfo = getimagesize($absolute_path);
-        if ($imageinfo['mime'] != 'image/gif' && $imageinfo['mime'] != 'image/jpeg' && $imageinfo['mime'] != 'image/jpg' && $imageinfo['mime'] != 'image/png') {
-            $this->logger->log('debug','processed image with bad header. Skipping '.$fullpath);
+        try {
+            $res = $this->client->get($absolutePath);
+        } catch (\Exception $e) {
+            $this->logger->error('DownloadImages: Can not retrieve image, skipping.', ['exception' => $e]);
 
             return false;
         }
 
-        // regenerate image
-        $this->logger->log('debug','regenerating image');
-        $im = imagecreatefromstring($rawdata);
-        if ($im === false) {
-            $this->logger->log('error','error while regenerating image '.$fullpath);
+        $ext = $this->getExtensionFromResponse($res, $imagePath);
+        if (false === $res) {
+            return false;
+        }
+
+        $hashImage = hash('crc32', $absolutePath);
+        $localPath = $folderPath . '/' . $hashImage . '.' . $ext;
+
+        try {
+            $im = imagecreatefromstring($res->getBody());
+        } catch (\Exception $e) {
+            $im = false;
+        }
+
+        if (false === $im) {
+            $this->logger->error('DownloadImages: Error while regenerating image', ['path' => $localPath]);
 
             return false;
         }
 
-        switch ($imageinfo['mime']) {
-            case 'image/gif':
-                $result = imagegif($im, $fullpath);
-                $this->logger->log('debug','Re-creating gif');
-                break;
-            case 'image/jpeg':
-            case 'image/jpg':
-                $result = imagejpeg($im, $fullpath, REGENERATE_PICTURES_QUALITY);
-                $this->logger->log('debug','Re-creating jpg');
+        switch ($ext) {
+            case 'gif':
+                imagegif($im, $localPath);
+                $this->logger->debug('DownloadImages: Re-creating gif');
                 break;
-            case 'image/png':
-                $this->logger->log('debug','Re-creating png');
-                $result = imagepng($im, $fullpath, ceil(REGENERATE_PICTURES_QUALITY / 100 * 9));
+            case 'jpeg':
+            case 'jpg':
+                imagejpeg($im, $localPath, self::REGENERATE_PICTURES_QUALITY);
+                $this->logger->debug('DownloadImages: Re-creating jpg');
                 break;
+            case 'png':
+                imagealphablending($im, false);
+                imagesavealpha($im, true);
+                imagepng($im, $localPath, ceil(self::REGENERATE_PICTURES_QUALITY / 100 * 9));
+                $this->logger->debug('DownloadImages: Re-creating png');
         }
+
         imagedestroy($im);
 
-        return $result;
+        return $this->wallabagUrl . '/assets/images/' . $relativePath . '/' . $hashImage . '.' . $ext;
     }
 
-    private static function getAbsoluteLink($relativeLink, $url)
+    /**
+     * Remove all images for the given entry id.
+     *
+     * @param int $entryId ID of the entry
+     */
+    public function removeImages($entryId)
     {
-        /* return if already absolute URL */
-        if (parse_url($relativeLink, PHP_URL_SCHEME) != '') {
-            return $relativeLink;
+        $relativePath = $this->getRelativePath($entryId);
+        $folderPath = $this->baseFolder . '/' . $relativePath;
+
+        $finder = new Finder();
+        $finder
+            ->files()
+            ->ignoreDotFiles(true)
+            ->in($folderPath);
+
+        foreach ($finder as $file) {
+            @unlink($file->getRealPath());
         }
 
-        /* queries and anchors */
-        if ($relativeLink[0] == '#' || $relativeLink[0] == '?') {
-            return $url.$relativeLink;
+        @rmdir($folderPath);
+    }
+
+    /**
+     * Get images urls from the srcset image attribute.
+     *
+     * @param Crawler $imagesCrawler
+     *
+     * @return array An array of urls
+     */
+    private function getSrcsetUrls(Crawler $imagesCrawler)
+    {
+        $urls = [];
+        $iterator = $imagesCrawler
+            ->getIterator();
+        while ($iterator->valid()) {
+            $srcsetAttribute = $iterator->current()->getAttribute('srcset');
+            if ('' !== $srcsetAttribute) {
+                // Couldn't start with " OR ' OR a white space
+                // Could be one or more white space
+                // Must be one or more digits followed by w OR x
+                $pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
+                preg_match_all($pattern, $srcsetAttribute, $matches);
+                $srcset = call_user_func_array('array_merge', $matches);
+                $srcsetUrls = array_map(function ($src) {
+                    return trim(explode(' ', $src, 2)[0]);
+                }, $srcset);
+                $urls = array_merge($srcsetUrls, $urls);
+            }
+            $iterator->next();
         }
 
-        /* parse base URL and convert to local variables:
-           $scheme, $host, $path */
-        extract(parse_url($url));
+        return $urls;
+    }
 
-        /* remove non-directory element from path */
-        $path = preg_replace('#/[^/]*$#', '', $path);
+    /**
+     * Setup base folder where all images are going to be saved.
+     */
+    private function setFolder()
+    {
+        // if folder doesn't exist, attempt to create one and store the folder name in property $folder
+        if (!file_exists($this->baseFolder)) {
+            mkdir($this->baseFolder, 0755, true);
+        }
+    }
+
+    /**
+     * Generate the folder where we are going to save images based on the entry url.
+     *
+     * @param int $entryId ID of the entry
+     *
+     * @return string
+     */
+    private function getRelativePath($entryId)
+    {
+        $hashId = hash('crc32', $entryId);
+        $relativePath = $hashId[0] . '/' . $hashId[1] . '/' . $hashId;
+        $folderPath = $this->baseFolder . '/' . $relativePath;
+
+        if (!file_exists($folderPath)) {
+            mkdir($folderPath, 0777, true);
+        }
+
+        $this->logger->debug('DownloadImages: Folder used for that Entry id', ['folder' => $folderPath, 'entryId' => $entryId]);
+
+        return $relativePath;
+    }
 
-        /* destroy path if relative url points to root */
-        if ($relativeLink[0] == '/') {
-            $path = '';
+    /**
+     * Make an $url absolute based on the $base.
+     *
+     * @see Graby->makeAbsoluteStr
+     *
+     * @param string $base Base url
+     * @param string $url  Url to make it absolute
+     *
+     * @return false|string
+     */
+    private function getAbsoluteLink($base, $url)
+    {
+        if (preg_match('!^https?://!i', $url)) {
+            // already absolute
+            return $url;
         }
 
-        /* dirty absolute URL */
-        $abs = $host.$path.'/'.$relativeLink;
+        $base = new \SimplePie_IRI($base);
+
+        // remove '//' in URL path (causes URLs not to resolve properly)
+        if (isset($base->ipath)) {
+            $base->ipath = preg_replace('!//+!', '/', $base->ipath);
+        }
 
-        /* replace '//' or '/./' or '/foo/../' with '/' */
-        $re = array('#(/\.?/)#', '#/(?!\.\.)[^/]+/\.\./#');
-        for ($n = 1; $n > 0; $abs = preg_replace($re, '/', $abs, -1, $n)) {
+        if ($absolute = \SimplePie_IRI::absolutize($base, $url)) {
+            return $absolute->get_uri();
+        }
+
+        $this->logger->error('DownloadImages: Can not make an absolute link', ['base' => $base, 'url' => $url]);
+
+        return false;
+    }
+
+    /**
+     * Retrieve and validate the extension from the response of the url of the image.
+     *
+     * @param Response $res       Guzzle Response
+     * @param string   $imagePath Path from the src image from the content (used for log only)
+     *
+     * @return string|false Extension name or false if validation failed
+     */
+    private function getExtensionFromResponse(Response $res, $imagePath)
+    {
+        $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
+        $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
+
+        // ok header doesn't have the extension, try a different way
+        if (empty($ext)) {
+            $types = [
+                'jpeg' => "\xFF\xD8\xFF",
+                'gif' => 'GIF',
+                'png' => "\x89\x50\x4e\x47\x0d\x0a",
+            ];
+            $bytes = substr((string) $res->getBody(), 0, 8);
+
+            foreach ($types as $type => $header) {
+                if (0 === strpos($bytes, $header)) {
+                    $ext = $type;
+                    break;
+                }
+            }
+
+            $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
+        }
+
+        if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
+
+            return false;
         }
 
-        /* absolute URL is ready! */
-        return $scheme.'://'.$abs;
+        return $ext;
     }
 }